Brain tumor detection and segmentation using deep learning.

OBJECTIVES: Brain tumor detection, classification and segmentation are challenging due to the heterogeneous nature of brain tumors. Different deep learning-based algorithms are available for object detection; however, the performance of detection algorithms on brain tumor data has not been widely explored. Therefore, we aim to compare different object detection algorithms (Faster R-CNN, YOLO & SSD) for brain tumor detection on MRI data. Furthermore, the best-performing detection network is paired with a 2D U-Net for pixel-wise segmentation of abnormal tumor cells. MATERIALS AND METHODS: The proposed model was evaluated on the Brain Tumor Figshare (BTF) dataset, and the best-performing detection network cascaded with 2D U-Net for pixel-wise segmentation of tumors. The best-performing detection network was also fine-tuned on BRATS 2018 data to detect and classify the glioma tumor. RESULTS: For the detection of three tumor types, YOLOv5 achieved the highest mAP of 89.5% on test data compared to other networks. For segmentation, YOLOv5 combined with 2D U-Net achieved a higher DSC compared to the 2D U-Net alone (DSC: YOLOv5 + 2D U-Net = 88.1%; 2D U-Net = 80.5%). The proposed method was compared with the existing detection and segmentation network i.e. Mask R-CNN and achieved a higher mAP (YOLOv5 + 2D U-Net = 89.5%; Mask R-CNN = 67%) and DSC (YOLOv5 + 2D U-Net = 88.1%; Mask R-CNN = 44.2%). CONCLUSION: In this work, we propose a deep-learning-based method for multi-class tumor detection, classification and segmentation that combines YOLOv5 with 2D U-Net. The results show that the proposed method not only detects different types of brain tumors accurately but also delineates the tumor region precisely within the detected bounding box.

Compressed SVD-based L + S model to reconstruct undersampled dynamic MRI data using parallel architecture.

BACKGROUND: Magnetic Resonance Imaging (MRI) is a highly demanded medical imaging system due to high resolution, large volumetric coverage, and ability to capture the dynamic and functional information of body organs e.g. cardiac MRI is employed to assess cardiac structure and evaluate blood flow dynamics through the cardiac valves. Long scan time is the main drawback of MRI, which makes it difficult for the patients to remain still during the scanning process. OBJECTIVE: By collecting fewer measurements, MRI scan time can be shortened, but this undersampling causes aliasing artifacts in the reconstructed images. Advanced image reconstruction algorithms have been used in literature to overcome these undersampling artifacts. These algorithms are computationally expensive and require a long time for reconstruction which makes them infeasible for real-time clinical applications e.g. cardiac MRI. However, exploiting the inherent parallelism in these algorithms can help to reduce their computation time. METHODS: Low-rank plus sparse (L+S) matrix decomposition model is a technique used in literature to reconstruct the highly undersampled dynamic MRI (dMRI) data at the expense of long reconstruction time. In this paper, Compressed Singular Value Decomposition (cSVD) model is used in L+S decomposition model (instead of conventional SVD) to reduce the reconstruction time. The results provide improved quality of the reconstructed images. Furthermore, it has been observed that cSVD and other parts of the L+S model possess highly parallel operations; therefore, a customized GPU based parallel architecture of the modified L+S model has been presented to further reduce the reconstruction time. RESULTS: Four cardiac MRI datasets (three different cardiac perfusion acquired from different patients and one cardiac cine data), each with different acceleration factors of 2, 6 and 8 are used for experiments in this paper. Experimental results demonstrate that using the proposed parallel architecture for the reconstruction of cardiac perfusion data provides a speed-up factor up to 19.15× (with memory latency) and 70.55× (without memory latency) in comparison to the conventional CPU reconstruction with no compromise on image quality. CONCLUSION: The proposed method is well-suited for real-time clinical applications, offering a substantial reduction in reconstruction time.

Accelerated Diffusion-Weighted MR Image Reconstruction Using Deep Neural Networks.

Under-sampling in diffusion-weighted imaging (DWI) decreases the scan time that helps to reduce off-resonance effects, geometric distortions, and susceptibility artifacts; however, it leads to under-sampling artifacts. In this paper, diffusion-weighted MR image (DWI-MR) reconstruction using deep learning (DWI U-Net) is proposed to recover artifact-free DW images from variable density highly under-sampled k-space data. Additionally, different optimizers, i.e., RMSProp, Adam, Adagrad, and Adadelta, have been investigated to choose the best optimizers for DWI U-Net. The reconstruction results are compared with the conventional Compressed Sensing (CS) reconstruction. The quality of the recovered images is assessed using mean artifact power (AP), mean root mean square error (RMSE), mean structural similarity index measure (SSIM), and mean apparent diffusion coefficient (ADC). The proposed method provides up to 61.1%, 60.0%, 30.4%, and 28.7% improvements in the mean AP value of the reconstructed images in our experiments with different optimizers, i.e., RMSProp, Adam, Adagrad, and Adadelta, respectively, as compared to the conventional CS at an acceleration factor of 6 (i.e., AF = 6). The results of DWI U-Net with the RMSProp, Adam, Adagrad, and Adadelta optimizers show 13.6%, 10.0%, 8.7%, and 8.74% improvements, respectively, in terms of mean SSIM with respect to the conventional CS at AF = 6. Also, the proposed technique shows 51.4%, 29.5%, 24.04%, and 18.0% improvements in terms of mean RMSE using the RMSProp, Adam, Adagrad, and Adadelta optimizers, respectively, with reference to the conventional CS at AF = 6. The results confirm that DWI U-Net performs better than the conventional CS reconstruction. Also, when comparing the different optimizers in DWI U-Net, RMSProp provides better results than the other optimizers.

Transfer learning in deep neural network-based receiver coil sensitivity map estimation.

INTRODUCTION: The success of parallel Magnetic Resonance Imaging algorithms like SENSitivity Encoding (SENSE) depends on an accurate estimation of the receiver coil sensitivity maps. Deep learning-based receiver coil sensitivity map estimation depends upon the size of training dataset and generalization capabilities of the trained neural network. When there is a mismatch between the training and testing datasets, retraining of the neural networks is required from a scratch which is costly and time consuming. MATERIALS AND METHODS: A transfer learning approach, i.e., end-to-end fine-tuning is proposed to address the data scarcity and generalization problems of deep learning-based receiver coil sensitivity map estimation. First, generalization capabilities of a pre-trained U-Net (initially trained on 1.5T receiver coil sensitivity maps) are thoroughly assessed for 3T receiver coil sensitivity map estimation. Later, end-to-end fine-tuning is performed on the pre-trained U-Net to estimate the 3T receiver coil sensitivity maps. RESULT AND CONCLUSION: Peak Signal-to-Noise Ratio, Root Mean Square Error and central line profiles (of the SENSE reconstructed images) show a successful SENSE reconstruction by utilizing the receiver coil sensitivity maps estimated by the proposed method.

Parallel implementation of L + S signal recovery in dynamic MRI.

Dynamic MRI is useful to diagnose different diseases, e.g. cardiac ailments, by monitoring the structure and function of the heart and blood flow through the valves. Faster data acquisition is highly desirable in dynamic MRI, but this may lead to aliasing artifacts due to under-sampling. Advanced image reconstruction algorithms are required to obtain aliasing-free MR images from the acquired under-sampled data. One major limitation of using the advanced reconstruction algorithms is their computationally expensive and time-consuming nature, which make them infeasible for clinical use, especially for applications like cardiac MRI. L + S decomposition model is an approach provided in literature which separates the sparse and low-rank information in dynamic MRI. However, L + S decomposition model is a computationally complex process demanding significant computation time. In this paper, a parallel framework is proposed to accelerate the image reconstruction process of L + S decomposition model using GPU. Experiments are performed on cardiac perfusion dataset ([Formula: see text]) and cardiac cine dataset ([Formula: see text]) using NVIDIA's GeForce GTX780 GPU and Core-i7 CPU. The results show that the proposed method provides up to 18 × speed-up including the memory transfer time (i.e. data transfer between the CPU and GPU) and ~ 46 × speed-up without memory transfer for the cardiac perfusion dataset in our experiments. This level of improvement in the reconstruction time will increase the usefulness of L + S reconstruction by making it feasible for clinical applications.

Respiratory motion-corrected, compressively sampled dynamic MR image reconstruction by exploiting multiple sparsity constraints and phase correlation-based data binning.

INTRODUCTION: Cardiac magnetic resonance imaging (cMRI) is a standard method that is clinically used to evaluate the function of the human heart. Respiratory motion during a cMRI scan causes blurring artefacts in the reconstructed images. In conventional MRI, breath holding is used to avoid respiratory motion artefacts, which may be difficult for cardiac patients. MATERIALS AND METHODS: This paper proposes a method in which phase correlation-based binning, followed by image registration-based sparsity along with spatio-temporal sparsity, is incorporated into the standard low rank + sparse (L+S) reconstruction for free-breathing cardiac cine MRI. The proposed method is validated on clinical data and simulated free-breathing cardiac cine data for different acceleration factors (AFs). The reconstructed images are analysed using visual assessment, artefact power (AP) and root-mean-square error (RMSE). The results of the proposed method are compared with the contemporary motion-corrected compressed sensing (MC-CS) method given in the literature. RESULTS: Our results show that the proposed method successfully reconstructs the motion-corrected images from respiratory motion-corrupted, compressively sampled cardiac cine MR data, e.g., there is 26% and 24% improvement in terms of AP and RMSE values, respectively, at AF = 4 and 20% and 16.04% improvement in terms of AP and RMSE values, respectively, at AF = 8 in the reconstruction results from the proposed method for the cardiac phantom cine data. CONCLUSION: The proposed method achieves significant improvement in the AP and RMSE values at different AFs for both the phantom and in vivo data.

Accelerating GRAPPA reconstruction using SoC design for real-time cardiac MRI.

Real-time cardiac MRI is a rapidly developing area of research that has the potential to improve the diagnosis and treatment of cardiovascular diseases. However, the acquisition of high-quality real-time cardiac MR (CMR) images is challenging as it requires a high frame rate and temporal resolution. To overcome this challenge, there have been recent efforts on several approaches including hardware-based improvements and image reconstruction techniques such as compressed sensing and parallel MRI. The use of parallel MRI techniques such as GRAPPA (Generalized Autocalibrating Partial Parallel Acquisition) is a promising approach for improving the temporal resolution of MRI and expanding its applications in clinical practice. However, the GRAPPA algorithm involves a significant amount of computation, particularly for high acceleration factors and large datasets. This can result in long reconstruction times, which can limit the ability to achieve real-time imaging or high frame rates. One solution to this challenge is to use specialized hardware i.e. field-programmable gate arrays (FPGAs). In this work, a novel 32-bit floating-point FPGA-based GRAPPA accelerator is proposed with an aim to reconstruct high-quality cardiac MR images at higher frame rates, making it well suited for real-time clinical applications. The proposed FPGA-based accelerator consists of custom-designed data processing units named as dedicated computational engines (DCEs) that allow for a continuous flow of data between the calibration and synthesis stages of GRAPPA reconstruction process. This greatly increases the throughput and reduces the latency of the overall proposed system. Moreover, a high-speed memory module (DDR4-SDRAM) is integrated with the proposed architecture to store the multi-coil MR data. An on-chip quad-core ARM Cortex-A53 processor is used to manage access control information required for data transfer between the DCEs and DDR4-SDRAM. The proposed accelerator is implemented on Xilinx Zynq UltraScale + MPSoC using high-level synthesis (HLS) and hardware descriptive language (HDL) with an aim to explore the trade-offs between the reconstruction time, resource utilization and design effort. Several experiments have been performed using in-vivo cardiac datasets i.e. 18-receiver coil and 30-receiver coil to evaluate the performance of the proposed accelerator. A comparison is performed with the contemporary CPU and GPU-based GRAPPA reconstruction methods in terms of reconstruction time, frames-per-second and reconstruction accuracy (RMSE and SNR). The results show that the proposed accelerator achieves speed-up factors up to 121× and 9× as compared to the contemporary CPU-based and GPU-based GRAPPA reconstruction methods, respectively. Moreover, it has been demonstrated that the proposed accelerator can achieve reconstruction rates of up to ∼27 frames-per-second while maintaining the visual quality of the reconstructed images.

GPU accelerated grouped magnetic resonance fingerprinting using clustering techniques.

Magnetic Resonance Fingerprinting (MRF) is a new quantitative technique of Magnetic Resonance Imaging (MRI). Conventionally, MRF requires sequential correlation of the acquired MRF signals with all the signals of (a large sized) MRF dictionary. This is a computationally intensive matching process and is a major challenge in MRF image reconstruction. This paper introduces the use of clustering techniques (to reduce the effective size of MRF dictionary) by splitting MRF dictionary into multiple small sized MRF dictionary components called MRF signal groups. The proposed method has been further optimized for parallel processing to reduce the computation time of MRF pattern matching. A multi-core GPU based parallel framework has been developed that enables the MRF algorithm to process multiple MRF signals simultaneously. Experiments have been performed on human head and phantom datasets. The results show that the proposed method accelerates the conventional MRF (MATLAB based) reconstruction time up to 25× with single-core CPU implementation, 300× with multi- core CPU implementation and 1035× with the proposed multi-core GPU based framework by keeping the SNR of the resulting images in a clinically acceptable range. Furthermore, experimental results show that the memory requirements of MRF dictionary get significantly reduced (due to efficient memory utilization) in the proposed method.

GPU accelerated Cartesian GRAPPA reconstruction using CUDA.

BACKGROUND AND OBJECTIVE: GRAPPA (Generalized Auto-calibrating Partially Parallel Acquisition) is an advanced parallel MRI reconstruction method (pMRI) that enables under-sampled data acquisition with multiple receiver coils to reduce the MRI scan time and reconstructs artifact free image from the acquired under-sampled data. However, the reduction in MRI scan time comes at the expense of long reconstruction time. It is because the GRAPPA reconstruction time shows exponential growth with increasing number of receiver coils. Consequently, the conventional CPU platforms may not adhere to the requirements of fast data processing for MR image reconstruction. METHODS: Graphics Processing Units (GPUs) have recently emerged as a viable commodity hardware to reduce the reconstruction time of pMRI methods. This paper presents a novel GPU based implementation of GRAPPA using custom built CUDA kernels, to meet the rising demands of fast MRI processing. The proposed framework exploits intrinsic parallelism in the calibration and synthesis phases of GRAPPA reconstruction process, aiming to achieve high speed MR image reconstruction for various GRAPPA configuration settings using different number of receiver coils, auto-calibration signals (ACS), sizes of GRAPPA kernel and acceleration factors. In-vivo experiments (using 8, 12 and 30 receiver coils) are performed to compare the performance of the proposed GPU accelerated GRAPPA with the CPU based GRAPPA extensions and GPU counterpart. RESULTS: The results indicate that the proposed method achieves up to ≈47.8× , ≈17× and ≈3.8× speed up gains over multicore CPU (single thread), multicore CPU (8 thread) and Gadgetron (GPU based GRAPPA) respectively, without compromising the reconstruction accuracy. CONCLUSIONS: The proposed method reduces the GRAPPA reconstruction time by employing the calibration phase (GRAPPA weights estimation) and synthesis phase (interpolation) on GPU. Our study shows that the proposed GPU based parallel framework for GRAPPA reconstruction provides a solution for high-speed image reconstruction while maintaining the quality of the reconstructed images.

Real-time, single breath-hold, multi-slice, 2D cine radial MR image reconstruction using sc-GROG k-t ESPIRiT.

Background:Multi-slice, multiple breath-hold ECG-gated 2D cine MRI is a standard technique for evaluating heart function and restricted to one or two images per breath-hold. Therefore, the standard cine MRI requires long scan time and can result in slice-misalignments because of various breath-hold locations in the multiple acquisitions.Methods:This work proposes the sc-GROG based k-t ESPIRiT with Total Variation (TV) constraint (sc-GROG k-t ESPIRiT) to reconstruct unaliased cardiac real-time cine MR images from highly accelerated whole heart multi-slice, single breath-hold, real-time 2D cine radial data acquired using the balanced steady-state free precession (trueFISP) sequence in 8 patients. The proposed method quality is assessed via Artifact Power (AP), Root-Mean Square Error (RMSE), Structure Similarity Index (SSIM), Peak Signal-to-Noise Ratio (PSNR), blood-pool to myocardial Contrast-to-Noise-Ratio (CNR), Signal-to-Noise-Ratio (SNR) and spatial-temporal intensity plots through the blood-myocardium boundary. The proposed method quantitative results are compared with the NUFFT based k-t ESPIRiT with Total Variation (TV) constraint (NUFFT k-t ESPIRiT) approach. Furthermore, clinical analysis and function quantification are assessed by Bland-Altman (BA) analyses.Results:As supported by the visual assessment and evaluation parameters, the reconstruction results of the sc-GROG k-t ESPIRiT approach provide an average 21%, 12%, 1% and 47% improvement in AP, RMSE, SSIM and PSNR, respectively in comparison to the NUFFT k-t ESPIRiT approach. Furthermore, the proposed method gives on average 45% and 58% improved blood-pool to myocardial CNR and SNR than the NUFFT k-t ESPIRiT approach. Also, from the BA plot, the proposed method gives better left ventricular and right ventricular function measurements as compared to the NUFFT k-t ESPIRiT scheme.Conclusions:The sc-GROG k-t ESPIRiT (Proposed Method) improves the spatio-temporal quality of the whole heart multi-slice, single breath-hold, real-time 2D cine radial MR and semi-automated analysis using standard clinical software, as compared to the NUFFT k-t ESPIRiT approach.